Humanoid robots in 2026: what they can actually do, and what is still hype
They are real, they are on factory floors, and the money behind them is staggering. They also run out of battery in two hours and need a human babysitter. A grounded look at where humanoid robots actually stand, past the demo reels.
Humanoid robots crossed a real line in 2026: they are now doing paid work on actual factory floors, not just demos. But the gap between that and the hype is enormous. By industry estimates only a couple hundred humanoids are doing genuine paid work worldwide, handling simple tasks like moving totes and loading parts, running out of battery in an hour or two, and needing human supervision. The money is real (Figure is valued around $39 billion, Tesla is starting Optimus production), the factory pilots are real, and the robot in your kitchen is still years away.
Every few years the humanoid robot has its moment, a slick demo where a machine folds laundry or pours a drink, and everyone declares the future has arrived. Then nothing ships. 2026 is different, but not in the way the demo reels suggest. The robots are genuinely working now, for money, in real buildings. They are also far narrower, slower and dumber than the marketing implies. Both of those things are true at once, and the space between them is the actual story.
Are humanoid robots actually real in 2026?
Yes, and this is the part that is genuinely new. Two-legged robots are clocking in for paid shifts on factory and warehouse floors, doing repetitive work alongside people. This is not a staged clip; it is a job.
The reality check comes with the numbers. By most estimates, still only a few hundred humanoid robots are doing real paid work in the entire world right now. Set that against the forecasts, Goldman Sachs sees more than a million humanoids in service by 2035, Bank of America has projected three billion of them worldwide by 2060, and you can see the size of the leap still to be made. The machines have crossed from demo to deployment. They have not crossed from deployment to scale.
What can they actually do?
Useful, boring, repetitive things, and not much beyond that yet.
The clearest proof point is Figure, whose robots have been working at a BMW plant in Spartanburg, South Carolina. In an eleven-month pilot its Figure 02 model logged more than 1,250 operating hours, loaded over 90,000 sheet-metal parts and helped build more than 30,000 vehicles, working ten-hour shifts at above 99% placement accuracy. Its newer Figure 03 has since taken over, adding sensor-packed hands and palm cameras. Separately, Agility Robotics' Digit is working commercial shifts moving totes at a Toyota plant in Canada and a GXO logistics warehouse in Georgia, and Amazon is using it in fulfilment.
Now the limits, because they matter. These robots move plastic bins, load sheet metal and carry parts a few feet to a conveyor. They typically run out of battery in an hour or two, often work behind safety barriers, and need a human to step in every few hundred hours. The genuinely hard, high-speed, high-precision factory jobs, welding, stamping, are still done by the bolted-down industrial arms that have existed for decades. The humanoid's advantage is not that it is better; it is that it is shaped like us, so it can slot into a workplace built for humans without rebuilding the workplace.
Who is actually building them?
The field is crowded and well funded. The names worth knowing:
| Maker | Robot | Where it stands |
|---|---|---|
| Figure (US) | Figure 03 | Deployed at BMW; valued around $39B, backed by Nvidia and Bezos |
| Tesla (US) | Optimus | Works in Tesla's own factories; low-volume production from 2026, high-volume targeted 2027 |
| Agility Robotics (US) | Digit | Paid warehouse shifts at Toyota, GXO and Amazon |
| Unitree (China) | G1 | Low-cost, sold widely; China is pushing hard on volume |
| 1X (Norway/US) | Neo | Aimed at the home, consumer pre-orders open |
The money tells the story: humanoid-specific funding is estimated to have reached several billion dollars in 2025, up from a few hundred million a few years earlier. Whatever you think of the timelines, the capital is treating this as real.
How much does a humanoid robot cost?
Less than you would guess, and that is the point. China's Unitree sells its G1 from roughly $13,500 (down from its $16,000 launch price), cheap enough to reshape the whole conversation. 1X opened consumer pre-orders for its home robot, Neo, at $20,000 outright or about $499 a month, with deliveries starting in 2026. Tesla's Elon Musk keeps pointing at a long-term $20,000 to $30,000 target for Optimus at scale, though that is a goal, not a price tag you can pay today.
Two cautions on all of this. Musk's timelines have a long history of slipping, most recently on Tesla's own robotaxi rollout, so treat "mass production in 2026" as an ambition. And a cheap home robot you can pre-order is not the same as a useful one you would actually want, which is a lesson the whole field is about to learn in public.
Will they take people's jobs?
This is the real question, and the honest answer in 2026 is "not yet, but that is not the same as never."
So far there are no documented layoffs directly attributable to humanoid robots, and the current deployments are aimed at jobs companies struggle to fill anyway: repetitive warehouse and logistics roles with high turnover and chronic worker shortages. For now the robots are plugging holes, not replacing people. But the direction is not subtle. Industry conferences increasingly treat labour displacement as a central theme, and China has made a national priority of moving humanoids from demos into real jobs across factories, logistics and even hospitals. The technology is being built, explicitly, to do work people currently do. Pretending otherwise helps no one.
So how close is the robot future?
Closer than the sceptics say, and much further than the demo reels imply. In a handful of factories and warehouses, the humanoid robot is a real, paid worker doing real, if simple, tasks, and that genuinely was not true two years ago. But it is a couple hundred machines doing narrow jobs on short battery life under human watch, not a workforce, and certainly not the helpful robot butler in the kitchen. The useful stance is the same one that fits self-driving cars: it is real where it is real, it is spreading, and the last, longest stretch, from "works in a controlled factory" to "works anywhere, unsupervised", has barely begun. For more from the edge, see the Future section and our look at robotaxis in 2026.
Quantum computing just had its 'this is a real industry now' week
In a few days in August, quantum computing companies posted triple-digit revenue growth, went live inside a big cloud platform, and landed government chip money. It is a genuine inflection, from lab curiosity to business. It is also not the same as quantum computers being broadly useful yet, and telling those two apart is the whole point.
Google's Pixel 11 is here, and the real product is Gemini everywhere
At Made by Google 2026, Google unveiled a full Pixel 11 line, a folding flagship, a new watch and its first Bluetooth tracker, all built on a 2nm Tensor G6 chip. But the hardware was the supporting act. The headline features are AI ones, a camera that turns sign language into text and a voice input that understands rambling speech, and the prices crept up about a hundred dollars. Here is what launched and what it signals.
Fusion power just got a $4 billion vote of confidence, and a date to prove itself
The old joke is that fusion is always 30 years away. In 2026 the private leaders are trying to make it a dated bet instead: Commonwealth Fusion just raised its total to $4 billion, its reactor is most of the way built, and it is aiming for net energy gain in 2027. Here is what is real, and the caveat that separates a lab result from a power plant.